基于FPGA的类脑神经元仿生电路设计与实现  

Design and implementation of a brain-like neural elementbionic circuit based on FPGA

作　　者：徐桂芝[1,2] 郭嘉荣 张慧港 郭磊[1,2] XU Gui-zhi;GUO Jia-rong;ZHANG Hui-gang;GUO Lei(State Key Laboratory of Reliability and Intelligence of Electrical Equipment,Hebei University of Technology,Tianjin 300130,China;Hebei Key Laboratory of Bioelectromagnetics and Neuroengineering,Hebei University of Technology,Tianjin 300130,China)

机构地区：[1]河北工业大学省部共建电工装备可靠性与智能化国家重点实验室,天津300130 [2]河北工业大学河北省生物电磁与神经工程重点实验室,天津300130

出　　处：《电机与控制学报》2023年第2期61-68,共8页Electric Machines and Control

基　　金：国家自然科学基金(51977060,51737003)。

摘　　要：随着现代社会进入信息化与智能化时代,神经形态计算以高度的并行、极低的功耗和存算一体的特征受到了广泛的关注。本文着眼于未来神经形态器件和类脑智能的发展需求,对建立全数字电路的类脑计算硬件系统展开研究,以脉冲神经网络H-H(Hodgkin-Huxley)神经元模型为研究对象,将DSP Builder平台与Quartus平台相结合,设计出H-H神经元模型的数字电路,并对其进行了性能测试。结果表明,基于现场可编程门阵列(FPGA)硬件设计的神经元数字电路其工作特性与理论仿真结果高度一致,具有良好的生物特性,为建立类脑计算硬件系统提供新思路。As modern society enters the era of information and intelligence,neuromorphic computing has attracted extensive attention for its high parallelism,low power consumption and integration of memory and computation.This paper focuses on the development needs of neuromorphic devices and brain-like intelligence in the future,and studies the brain-like computing hardware system of the whole digital circuit.Taking H-H(Hodgkin-Huxley)neuron model of pulse neural network as the research object,the digital circuit of H-H neuron model was designed by combining DSP Builder platform and Quartus platform.And the performance test was carried out.The results show that the working characteristics of the neuron digital circuit based on field programmable gate array(FPGA)hardware design are highly consistent with the theoretical simulation results,and have good biological characteristics,which provides a new idea for the establishment of brain-like computing hardware syste m.

关 键 词：类脑计算 脉冲神经网络 现场可编程门阵列 Quartus H-H神经元模型 DSP Builder 

分 类 号：TM935[电气工程—电力电子与电力传动] Q811.21[生物学—生物工程]